Investigators reviewed records from the Arkansas Poison Control Center for the period of May 1, 2021, through July 25, 2023, screening all cases where the recorded exposure was snake envenomation. Reviewers included cases where crotaline antivenom was administered and excluded cases where (1) antivenom was not administered or its administration was uncertain, (2) the type of antivenom was unknown, or (3) a hypersensitivity reaction preceded antivenom administration (Figure 1).

The available demographic data from the Arkansas Poison Control Center included age and sex. The narrative data provided information on the type of antivenom administered, the number of vials used for treatment, the time from the bite to treatment (in hours), the species of snake responsible for envenomation, the bite location, the hospital length of stay (LOS, in hours), symptoms associated with envenomation, serum IgE levels to α-GAL (n=5), and any ADRs attributed to the antivenom. The description of specific ADR symptoms (e.g., hypotension, wheezing, edema, etc.) was abstracted from the narratives and not based on pre-specified criteria. A single individual obtained these data from the narratives.

Two reviewers determined the presence of ADRs in an independent, unblinded review of the narrative abstractions obtained from above. The two initial reviewers conducted preliminary scoring by consensus, categorizing cases as either ADR or no ADR, and as presumed anaphylaxis or no presumed anaphylaxis.

Cases identified as presumed anaphylaxis during the initial review underwent independent secondary review by two physician allergists blinded to the type of antivenom given and the prior reviewers’ label of presumed anaphylaxis. These reviewerspresumed anaphylaxis based on criterion 2 of the NIAID/FAAN guidelines, which was agreed upon before the review of cases occurred:^{25, 26}

We labeled these reactions as presumed anaphylaxis because we could not confirm whether IgE-mediated responses (snake antivenom or α-GAL) caused them. The blinded and unblinded reviewers reached a consensus on the final data coding.

We focused on the frequency of ADRs related to snake antivenoms in Arkansas and whether the incidence of ADRs differed between antivenom formulations. We report the α-GAL IgE concentrations (ng/dL) of those subjects who had these values recorded in the Poison Control Center narratives.

A Cohen’s kappa statistic was performed to determine inter-rater agreement for both the unblinded and blinded reviews. Descriptive statistics are presented as medians with interquartile ranges (IQR, 25th–75th percentiles) or means with standard deviations (SD) for continuous variables and as frequencies with percentages for discrete variables. We compared the difference (delta) with 95% confidence intervals (CI) for the ADRs and presumed anaphylaxis between those who received F(ab’)₂ versus Fab. We also compared the difference (delta) with 95% confidence intervals (CI) for the ADRs and presumed anaphylaxis between those who received F(ab’)₂ versus Fab.

Peripheral blood samples were collected from patients at the University of Virginia (UVA) with history of mammalian meat allergy and alpha-gal IgE sensitization in an IRB-approved study.²⁷ Five α-GAL allergic subjects and one non-allergic control were investigated. UVA had previously analyzed the plasma from these subjects for specific IgE to α-GAL using the ImmunoCAP 250 instrument (ThermoFisher/Phadia Scientific). None of the participants had received F(ab’)₂ or Fab antivenoms.

The presence of IgE in the serum indicates sensitization to the antigen, but it does not prove an allergic reaction will occur. Physiologically, a patient must have this specific IgE bound to mast cells and basophils in order for an allergic reaction to occur. To evaluate whether snake antivenom can stimulate a reaction in cells with IgE to α-gal bound to the high-affinity IgE receptor, we conducted basophil activation testing (BAT) on whole blood from α-GAL allergic and non-allergic (control) subjects using the Flow Cast kit (Bühlmann Diagnostics Corp.).²⁸ We performed BAT using established protocols²⁷ and measured the activation of basophils using flow cytometry, represented by increased CD63 expression. In these studies, a shift of the activation curve to the left indicates basophil activation at lower antigen concentrations. We tested dose-response effects with serial dilutions of snake antivenoms and appropriate controls (Supplemental Methods).

We compared the EC₅₀ for CD63 activation on basophils of the allergic and non-allergic subjects upon activation with different snake antivenoms and appropriate controls.

Dose-response curves and EC₅₀ values were generated using nonlinear regression and a three-parameter response. We compared the mean and standard deviation [SD] of these EC₅₀ values.

We used GraphPad Prism Software (version 10.2.3 for Windows; Boston, MA), R studio (v. 2024.09.1+394; Boston, MA), ²⁹ R (v. 4.4.2, Vienna, Austria), ³⁰ and Microsoft Excel for Mac (v. 16.88) to perform statistical analyses and create the figures and tables.

There were 439 snake bite records for which narrative abstraction was obtained; 218 of these cases were excluded because antivenom was not administered (n=199), there was unclear administration or unknown formulation of antivenom (n=18), or there was an antecedent allergic reaction (n=1). Fab antivenom alone was administered in 171 patient encounters, while F(ab’)₂ antivenom alone was administered in 37 encounters. Thirteen subjects received both antivenoms (Figure 1).

Most snake bites in this study were from the genus Agkistrodon and usually occurred on the hand or foot (Table 1 and Supplemental Table 1). As expected, more vials of F(ab’)₂ were administered compared to Fab because of the higher starting dose of F(ab’)₂ (Fab, mean [SD], 5.7 [2.7]; F(ab’)₂, 10.3 [2.04]). However, subjects received more Fab antivenom by weight compared to F(ab’)₂ antivenom (Fab, mean [SD], 5769mg [2662]; F(ab’)₂, 1238mg [235]), considering the greater weight of Fab per vial (1000mg/vial of Fab vs. 120mg/vial of F(ab’)₂).

Although there was perfect agreement (Cohen’s kappa = 1.0) regarding the presence of an ADR, agreement on reaction severity (presumed anaphylaxis vs. other ADR) was lower at κ = 0.69 (0.42 to 0.95) between the unblinded reviewers. The consensus achieved following the primary review matched the final set provided by the blinded Allergy and Immunology specialists (κ = 1.00).

Eight patients (21.6%) receiving F(ab’)₂ antivenom experienced ADR excluding presumed anaphylaxis, while seven patients (5.9%) receiving Fab antivenom (delta for F(ab’)₂ and Fab: 17.5% 95% CI [4.0–31.0]) (Table 1). Six patients (16.2%) receiving F(ab’)₂ antivenom developed presumed anaphylaxis compared with three patients (1.9%) receiving Fab antivenom (delta for F(ab’)₂ and Fab: 14.3% [2.5–26.2]) (Table 2). The distribution of the symptoms experienced and described as ADRs after either Fab, F(ab’)₂, or both is shown in Supplemental Figure 1. More subjects with an ADR to F(ab’)₂ had skin and/or mucosal symptoms (pruritis, rash, urticaria) and bronchospasm. Further, more than one body system was affected more often in subjects with an ADR to F(ab’)₂ than Fab, as reflected by the data regarding presumed anaphylaxis.

Five subjects with ADRs to antivenom had available α-GAL titers (Table 3), four of whom had received F(ab’)₂ and one Fab. Among these subjects, one had an existing red meat allergy, as determined from the Poison Center narrative abstraction, that had led to α-GAL testing. This subject had a known α-GAL IgE of 9.3 IU/mL (reference range < 0.10 IU/mL) measured seven years before the sustained presumed anaphylactic reaction to F(ab’)₂. Among the four other subjects with presumed anaphylaxis to F(ab’)₂ were a 65-year-old male and a 42-year-old female. Each had elevated α-GAL IgE titers: 4.4 and 4.3 kU/L, respectively. One subject had presumed anaphylaxis to Fab, a six-year-old patient with α-GAL IgE titer 47.00 IU/mL. The final patient, a 35-year-old male, sustained a cutaneous hypersensitivity response (not presumed anaphylaxis) to F(ab’)₂ and had a positive α-GAL titer of 0.56 IU/mL.

Supplemental Table 2 provides the demographics and IgE to α-GAL of subjects who provided blood samples for BAT.

In BAT testing, the dose-response curve shifts leftward as basophils become more readily activated by a potential allergen (Figure 2). The left shift demonstrates that a lower antigen concentration is required to stimulate the activation of the basophils, which is measured by higher levels of CD63. All subjects responded to the anti-FCeRI positive control (mean percentage of CD63 activation: 82.7% SD [14.73], n=6), and as expected, all the basophils from subjects with positive α-GAL reacted to beef thyroglobulin (mean percentage CD63 activation: 81.1% [20.74], n=5).

The basophils from four of the subjects with a known positive α-GAL test reacted to both types of antivenom. One responded to F(ab’)₂ but not Fab across the measured range of concentrations. The F(ab’)₂ antivenom was more potent than the Fab antivenom, but neither F(ab’)₂ nor Fab antivenoms were as potent as beef thyroglobulin, a positive control. In those that responded to both F(ab’)₂ and Fab, the F(ab’)₂ antivenom was 144 times more potent than the Fab antivenom based on the concentration of the antivenom (EC₅₀) required for activation of the cells (F(ab’)₂, mean [SD], 4.48 μg/mL [2.75]; Fab, 647 μg/mL [627]) (Figure 2). As expected, the control sample had no IgE to α-GAL and did not exhibit any basophil activation to beef thyroglobulin, F(ab’)₂, or Fab.